In the structural analysis Gepasi now also finds the elementary modes of the pathway, according to the algorithm of Schuster et al.
Gepasi is now capable of maximising or minimising any model variable, including functions of variables. This is implemented in a modular framework that allows the user to repeat the process with a series of optimisation methods.
6 DLLs implementing 6 optimisation methods are distributed with the program:
Steepest descent
L-BFGS-B (quasi-Newton method)
Levenberg-Marquardt (quasi-Newton method)
Hooke and Jeeves (direct search method)
random search
genetic algorithm
Further DLLs will be made available separatly that can be "plugged in" the current version of the program.
Data fitting now available suitable for steady state and time course experimental data. Advanced (but essential) features such as variable initial concentrations per experiement and the possibility of using any optimisation method for the fitting. Gepasi also allows multiple X (parameters) and Y (variables) columns for the fit.
Real and imaginary parts of the eigenvalues are now available for output on the data files (previously only on the report file). Two new eigenvalue statistics are calculated: stiffness (largest absolute real part divided by smallest absolute real part) and time hierarchy as defined by Heinrich and Schuster.
The simulation engine can now be paused from the toolbar (previously only from the time course page)
BUGS FIXED
A bug in the front end allowed the user to define an empty function (i.e. one without any definition). This would later cause an access violation and the program would terminate. Empty functions are no longer allowed to be defined.